OBJECTIVE: Fetal urine production at different gestational ages has been evaluated using ultrasound in several previous studies. In a recent study, we investigated the accuracy when estimating the bladder volume using the conventional ultrasound technique and found a total variability of 17.3-10.9% for bladder volumes of 5-40 mL. The variability is mainly caused by: (i) inappropriate image selection (the 'freezing error') and (ii) limitations when measuring on the frozen image (the 'frozen error'). The aim of this study was to reduce the total error by reducing the 'freezing' and the 'frozen error'. To this end, we used a modified manual ultrasound technique (adding a 'rocking' motion to the conventional method) and digitized the selected image. METHODS: Two patients for each gestational week from 24 to 40 weeks were selected. The fetal urinary bladder was examined with ultrasound three times within 1 min and documented on videotape. The volume, as assessed by the longitudinal section of the recorded bladder images, stored in digitized form, was evaluated on three occasions with > 24 h in between. The mean and variability (standard deviation, SD) were estimated. RESULTS: For fetal bladder volumes between 5 and 40 mL, the 'freezing error' (SD), the 'frozen error' and the 'total error' were 11.7-5.1%, 8.0-3.0% and 14.2-5.9%, respectively. Comparing the present with a previous study, when selecting images and assessing bladder volumes repeatedly within 1 min, SD was 12.9-5.5% vs. 17.3-10.9%. CONCLUSIONS: Using a modified ultrasound technique, the variability in fetal bladder volume estimation can be reduced.